Abstract
Abstract 406: Radiomics‐based Clot Profiling for Predicting Mechanical Thrombectomy Outcomes
Stroke: vascular and interventional neurology, Vol.5
11/01/2025
DOI: 10.1161/svi270000_406
PMCID: PMC12850364
Abstract
Clot composition might affect mechanical thrombectomy (MT) outcomes. Radiomics is a voxel‐based imaging technique that quantifies signal intensity (SI) through radiomic features (RFs). We aimed to characterize clot composition using radiomics in non‐contrast computed tomography (NCCT) scans to assess MT outcomes. Methods In Phase 1, we prospectively analyzed thrombi retrieved after MT. Samples were imaged with micro‐computed tomography (micro‐CT) and examined histologically. Micro‐CT and histological slices were matched to identify red blood cell (RBC)‐ and fibrin‐rich regions. RFs were extracted, multivariate logistic regression identified those associated with each component, and Spearman's rank correlation assessed RFs and histological percentage composition. The same clots were localized on NCCT, RFs were extracted, and correlations between micro‐CT and NCCT RFs were tested. Receiver operating characteristic analysis evaluated NCCT RFs for discrimination of clot composition. In Phase 2, radiomics were applied to a retrospective NCCT dataset of MT patients. Multivariate logistic regression was performed to address MT outcomes. Primary efficacy outcome was first‐pass effect (FPE), defined as modified treatment in cerebral infarction (mTICI) ≥2c on first attempt. Secondary efficacy outcomes included final excellent recanalization (mTICI ≥2c), failed recanalization (mTICI 0), and rescue stent placement. Safety outcomes included distal embolization and in‐hospital death. Results Ten thrombi were analyzed with micro‐CT. Total Energy (TE, OR 1.35, 95% CI 1.20‐1.54, p < 0.001) and Large Dependence High Gray Level Emphasis (LDHGLE, OR 1.18, 95% CI 1.07‐1.32, p = 0.01) were associated with RBCs and strongly correlated between micro‐CT and NCCT of clots with ≥70% of RBCs (Rho 0.752 and 0.815). In 250 retrospective NCCTs, TE and LDHGLE classified 150/250 (60%) as RBC‐rich and 100/250 (40%) as non‐RBC‐rich. RBC‐rich clots were more likely to achieve FPE (OR 3.2, 95% CI 1.84‐5.59, p < 0.001) and excellent recanalization (OR 2.6, 95% CI 1.53‐4.39, p < 0.001), and less likely to result in failed recanalization (OR 0.31, 95% CI 0.10‐0.94, p = 0.03) or require rescue stenting (OR 0.42, 95% CI 0.22‐0.78, p < 0.001). No associations were found with distal embolization (OR 0.72, 95% CI 0.39‐1.35, p = 0.73) or in‐hospital death (OR 0.60, 95% CI 0.33‐1.07, p = 0.08). ConclusionsRadiomics is a promising tool to assess MT outcomes considering clot composition.
Details
- Title: Subtitle
- Abstract 406: Radiomics‐based Clot Profiling for Predicting Mechanical Thrombectomy Outcomes
- Creators
- A. S Gudino - University of IowaA Van Dam - University of IowaM CabariqueR JaramilloE Baniewicz - University of IowaC Idrovo - University of Iowa, NeurologyL Diaz - University of Iowa, NeurologyC Dier - University of IowaN Shenoy - University of IowaC Aamot - University of IowaS Walsh - University of IowaO Lopes Abath Neto - University of IowaM Ghannam - University of IowaE Samaniego - University of Iowa
- Resource Type
- Abstract
- Publication Details
- Stroke: vascular and interventional neurology, Vol.5
- DOI
- 10.1161/svi270000_406
- PMCID
- PMC12850364
- ISSN
- 2694-5746
- eISSN
- 2694-5746
- Publisher
- Wiley Subscription Services, Inc
- Language
- English
- Date published
- 11/01/2025
- Academic Unit
- Neurology; Radiology; Pathology; Iowa Neuroscience Institute; Neurosurgery
- Record Identifier
- 9985089132902771
Metrics
8 Record Views